BACKGROUND: An elevated international normalized ratio (INR) increases the risk for major hemorrhage during warfarin therapy. Optimal management of patients with asymptomatic elevations in INR is hampered by the lack of understanding of the time course of INR decay after cessation of warfarin therapy. OBJECTIVE: To identify predictors of the rate of INR normalization after excessive anticoagulation. DESIGN: Retrospective cohort study. SETTING: Outpatient anticoagulant therapy unit. PATIENTS: Outpatients with an INR greater than 6.0 were identified from August 1993 to September 1998. Patients in whom two doses of warfarin were withheld and a follow-up INR was obtained on the second calendar day were enrolled. No patient received vitamin K(1). MEASUREMENTS: The INR was measured 2 days after an INR greater than 6.0 was recorded. RESULTS: Of 633 study patients with an initial INR greater than 6.0, 232 (37%) still had an INR of 4.0 or greater after two doses of warfarin were withheld. Patients who required larger weekly maintenance doses of warfarin were less likely to have an INR of 4.0 or greater on day 2 (adjusted odds ratio per 10 mg of warfarin, 0.87 [95% CI, 0.79 to 0.97]). Other risk factors for having an INR of 4.0 or greater on day 2 included age (odds ratio per decade of life, 1.18 [CI, 1.01 to 1.38]), index INR (odds ratio per unit, 1.25 [CI, 1.14 to 1.37]), decompensated congestive heart failure (odds ratio, 2.79 [CI, 1.30 to 5.98]), and active cancer (odds ratio, 2.48 [CI, 1.11 to 5.57]). CONCLUSIONS: Steady-state warfarin dose, advanced age, and extreme elevation in INR are risk factors for prolonged delay in return of the INR to within the therapeutic range. Decompensated congestive heart failure and active cancer greatly increase this risk.